D-Wave entangles dual-rail qubits preserving error hierarchy

D-Wave has entangled two dual-rail qubits without disrupting their natural error hierarchy, the company reported in Nature on August 5, 2026. The milestone advances gate-based quantum computing using a qubit design that could simplify error correction.

Dual-rail qubits store a single photon in a superposition of two resonators. Their most frequent error, photon escape, is detectable without extra qubits. This skewed error profile may allow error correction with fewer physical qubits per logical qubit, reducing hardware demands.

Trevor Lanting, D-Wave’s vice president of quantum technology, told Ars Technica that two-qubit gates often distort the error hierarchy, but the new gate maintains it: photon loss remains dominant, phase flips are less common, and bit flips are extremely rare. The operation completes in 500 nanoseconds, with 200 nanoseconds of actual interaction.

Experiments showed photon loss at roughly 0.5 percent per entanglement, five times more frequent than any other error. Bit flips measured at just 10^-6. Losses mostly affected the source qubit, which partially enters the tunable coupler that mediates the interaction.

D-Wave gained dual-rail expertise through its acquisition of Quantum Circuits, a startup whose technology is also used by Amazon. While Amazon connects dual-rail qubits via transmons, D-Wave entangles them directly. The company aims to assemble 181 dual-rail qubits by 2028 for surface code tests.

Challenges remain. Error rates increased quadratically with more gates, possibly from calibration drift or frequency fluctuations. Mid-circuit erasure detection is still under development, though Lanting said the company has a roadmap. Theoretical work on error decoding for this skewed error model is also needed.